Mendix 11 sharpens the everyday craft of building logic – a smoother microflow and nanoflow editing experience, tighter handling of return values and parameters, richer OQL and XPath querying, and more precise numbers and strings at runtime.
Working in the logic editors
The microflow and nanoflow editors are more comfortable to model in. Holding Shift while dragging disables snapping for fine placement, the border of a loop highlights while you drag over it, and connection points and resize handles appear only on the elements you have selected to reduce visual clutter. As you resize or insert elements, the editor auto-repositions neighbours and moves them out of the way – expanding a loop pushes connected objects aside, and resizing an element moves connected annotations along with everything else. You can also navigate context menus with the arrow keys in both the page and microflow editors.
Parameters and entity usage are clearer at a glance. Optional parameters are drawn with a dotted border so you can tell required from optional inputs immediately, and finding the usages of an entity now includes Change Object activities, giving a more complete picture of where that entity is touched.
- Automatic variable-to-parameter mapping – page, snippet, and action variables automatically map to microflow and nanoflow parameters that share the same name, cutting down on manual wiring.
- LogicBot recommender – the recommender dialog closes correctly when suggestions are cleared, rather than lingering after the app runs.
Calling flows, and using their results
Pages and snippets can capture and use the return values of microflows and nanoflows directly. Through an action return-value mapping, a primitive return can be assigned to a matching attribute or variable, and an object return can be mapped to multiple variables with optional expression transformation – no more workaround entities. This is handy for updating conditional visibility, setting filter variables after a calculation, or dynamically changing form behaviour based on validation.
Nanoflows can share list parameters by reference. A nanoflow-call setting lets a list be passed either as an isolated copy or as a shared reference; a shared reference syncs changes back to the caller the way microflows already do, making nanoflow list handling consistent with microflow behaviour.
- Client-side file saving – the modern Client API supports saving files by calling saveFile from mx-api/data, extending its data-handling capabilities for client-side logic.
- Following associations from variables in XPath – nanoflow database retrieves can follow an association from a variable, enabling comparisons such as $var/Association = $var.
Java and JavaScript actions
Java actions can mark Entity and Microflow parameters as optional, removing workarounds like dummy microflows or extra actions that previously existed only to satisfy required parameters. The Java API also gains flexibility through FeedbackHelper.addOpenFormFeedback, which can now open pages that have primitive and optional parameters when returning form-open feedback from a Java action.
- JavaScript Action Editor – a modernized editor for JavaScript actions is available in public beta and can be enabled under Preferences > New features.
Consuming REST services in a microflow
Wiring a REST integration into logic is faster: drag a consumed REST service directly onto a microflow and Mendix creates a Send REST Request activity for you. Refreshing the parameters in that activity’s Parameters grid now correctly updates the Base URL, calling an external action without arguments sends {} in the request body rather than an empty body, and opening a Studio Pro 10 app that uses the Send REST Request activity in Studio Pro 11 no longer raises a spurious consistency error for operations with query parameters.
Numbers and strings at runtime
Decimal precision is configurable per app. The default is 20 digits total with a scale of 8, and the scale – the digits after the decimal point – can be raised as high as 18, with total precision equal to the scale plus 20 (up to 20 digits before and 18 after the decimal). This serves financial and other domains that need extended precision; note that lowering the scale may require data migration and can cause data loss.
String null handling is a runtime choice. A Use new string behavior option on the Runtime tab controls how null or undefined values are treated for string attributes. The new behaviour – the default for new projects – keeps such values as null so nanoflows and microflows stay consistent, while the old behaviour converts them to empty strings and remains available for backward compatibility.
Querying with OQL
OQL is more expressive and easier to read. Queries support single-line comments with — and nesting block comments with /* … */, with Ctrl+/ (Cmd+/ on macOS) for line comments and Shift+Alt+A (Shift+Option+A on macOS) for block comments. The INSERT INTO statement can insert associations as well as attributes and can insert multiple objects with their associations. New functions round out the language: DATEFORMAT for formatting dates, and RPAD and LPAD for right- and left-padding strings.
- Grammar and semantics – OQL reserves additional keywords (including DELETE, INSERT, UPDATE, UPSERT, MERGE, and USING), and CAST accepts empty or NULL values to DATETIME, DECIMAL, INTEGER, LONG, and STRING. A UNION ORDER BY must reference only aliased columns, COALESCE follows numeric precedence (Decimal > Long > Integer), and dots are not allowed in aliases.
Querying with XPath
XPath constraints are stricter and more correct. Comparing an association or associated object to a Boolean or to empty – for example [MyModule.MyAssociation = false()] or [MyModule.MyAssociation = empty] – now produces a consistency error instead of being silently allowed, and the keywords true and false are treated the same as their function forms true() and false(). For safety, mathematical operators are disabled by default in Java XPath queries; code that relies on them must explicitly call allowMathOperators(true).
- Cleaner modeling feedback – the false-positive consistency error previously raised for query parameters in microflow activities when Included is set to No has been removed.
Workflows and BPMN. Mendix 11 turns the workflow editor into a far more expressive, BPMN-aligned canvas – richer event handling, multiple ways to conclude and interrupt a process, group-based task assignment, and AI agents that participate as first-class steps. Together these let you model complex, long-running business processes that stay correct as your organization and your model change over time.
Event handling and BPMN alignment
Boundary events attach to workflow activities to build timeout, escalation, and exception patterns directly in the model. Interrupting boundary events let a workflow abandon its modeled flow when a timer expires and branch into an alternate flow, while non-interrupting boundary events can be configured with an initial execution time, a recurrence count, and intervals ranging from minutes to months – ideal for repeated reminders and periodic checks within a running instance. Timer behavior is well-defined at the edges too: a standalone timer with a null or empty expression moves straight to Finished, and a boundary-event timer with an empty expression is never started and is marked Aborted, so you can model conditional timers cleanly.
Notification events bring BPMN message semantics into workflows so a process can pause and wait for a specific external signal – payment confirmations, API callbacks, and similar handoffs. Intermediate Message Events appear inline as a circle and act as a lighter alternative to the full Wait for Notification activity, providing intelligent waiting points. Notification Boundary Events attach to an activity and are triggered through the Notify Workflow microflow activity; interrupting variants cancel the current activity, while non-interrupting variants branch a parallel path without cancelling it.
Event sub-processes add non-interrupting, BPMN-style sub-processes that can be triggered at any moment during the workflow lifecycle to handle exceptions. Unlike boundary events, they run independently of any specific task, as optional process flows that do not disturb the main workflow. A notification event sub-process Start event can also be set to interrupting behavior, in which case it terminates all active execution paths within the global scope – giving you proper interrupt and cancel semantics for the whole workflow.
Multiple workflow endpoints remove the constraint of funneling every path into a single End activity. Different paths can now conclude independently through their own end events, user tasks and other multi-outcome activities can each lead to a distinct end event, and final elements can be jump activities – making complex business logic much clearer to read.
- Boundary event type switching – change a boundary event’s type directly from its context menu, without rebuilding the element.
Task assignment and instance management
Workflow groups make user assignment dynamic. Instead of targeting named individuals, you assign a user task to an entire group; anyone in that group automatically receives the task, and when people join or leave, assignments and access update instantly – so tasks stay correctly routed as team structures change. Groups can be modified at runtime without affecting already-initiated tasks, and the Workflow Commons module includes a management page for administering group membership.
Workflow conflict resolution, also part of Workflow Commons, lets you deal with instances that became incompatible after a model change. Rather than handling each stuck instance one at a time, incompatible instances are grouped so you can apply batch actions – Continue, Jump to Activity, Restart, or Abort – across the whole group.
- WorkflowEndedUserTask system entity – a System module entity records user tasks that have ended and distinguishes finished versus aborted outcomes, improving auditability and reporting on completed tasks.
- Robust multi-user tasks – a multi-user task with no targeted users now fails correctly instead of being left in an inconsistent state, and timer activities no longer get stuck when their workflow is upgraded while an instance is still active.
Intelligent and cross-application workflows
AI agents as workflow steps let you embed custom Mendix agents directly into workflow definitions as first-class citizens for AI-powered logic and decision-making. A dedicated AI Agent Task element in the workflow toolbox drops these AI-driven steps straight into the modeled flow.
The Global Inbox module centralizes workflow tasks from multiple applications into one place. Publisher applications emit events through the Global Inbox Connector, and Mendix Business Events keep the central inbox synchronized in real time, with the initial release focused on Mendix Workflows.
- Dynamic Case Management add-on – the CaseFabric Case Management add-on brings case models into Studio Pro using domain-specific languages for HumanTasks (UI), Process Tasks (logic), and case file structure via Mendix entities, and ships with a Dynamic Case Management Commons module of preconfigured templates and pages.
Modeling and tooling
- Find Usages for notifications – Find Usages identifies which Notify Workflow microflow activities target Wait for Notification activities and event sub-process start events, so you can trace how signals connect across a workflow.
- Safer editing – event sub-process and main start activities are no longer draggable, preventing accidental progress loss while you model.
- Debugger clarity – the Debugger pane adds document-type icons that make workflow elements easier to identify.
- Faster versioning – conflict analysis is quicker for workflows containing many nested outcome-based activities, reducing slowdowns during merge and version operations on complex models.